Intel: Winning and Losing

Original Article ↗ Hacker News Discussion ↗

Intel’s Strategic Missteps (Era & Scope)

  • Thread agrees the 2008–2014 focus misses the actual loss of dominance, which many place around:
    • 14nm stagnation and weak post‑Haswell generations.
    • The 2017–2019 rise of Ryzen and Apple’s switch to Apple Silicon.
  • Some readers find the article too spec‑sheet‑oriented and lacking deeper analysis of why Intel stumbled.

Atom, Quark/Edison, and Netbooks

  • Atom is widely remembered as a reputational breaking point: very slow laptops and “desperation” products.
  • Counterpoint: early Atom had decent perf‑per‑watt and enabled the netbook/nettop category; its DNA lives on in modern E‑cores.
  • Intel’s Quark/Edison line is seen as baffling: poor performance, worse efficiency and high BOM cost for embedded/IoT versus ARM SoCs.

Missing Mobile: XScale, ARM, and the Smartphone Era

  • Strong view that selling XScale and betting fully on x86 was a pivotal strategic error right before the smartphone boom.
  • Intel reportedly believed x86 and its process lead could win every segment; hindsight frames this as classic Innovator’s Dilemma and margin‑protection myopia.
  • Some note Atom’s CPU cores were not inherently inefficient, but were paired with power‑hungry chipsets, possibly to protect margins or cannibalization.

Itanium and Market Power

  • Mixed views:
    • Technically a failure, but credited by some with helping drive most non‑x86 server competitors out (except IBM).
    • Others argue plain x86 + Linux, not Itanium, killed 90s Unix workstations/servers; Apple also absorbed some workstation niches.
    • AMD is praised for preserving affordable x86 via x86‑64 and DDR, forcing Intel’s u‑turn.

Culture, Management, and Acquisitions

  • McAfee and other 2000s acquisitions are cited as evidence of a “finger in many pies” strategy that rarely produced wins.
  • Several commenters blame:
    • MBA/Wall‑Street mindset, stock buybacks, labor arbitrage, and layoffs during peak dominance.
    • Overgrown bureaucracy where telling leaders they’re wrong isn’t rewarded.
    • A comfortable 9‑to‑5 culture and risk‑averse protection of cash cows, stifling internal disruption and vision.

ISA, x86 vs ARM, and Performance

  • One camp: Intel over‑sold “ISA doesn’t matter,” then believed its own myth, underestimating architectural limits and ARM’s potential.
  • Another camp: x86 “tax” is real but small; modern OoO microarchitectures and memory‑latency hiding dominated performance, letting x86 crush most RISC in the 90s–2000s.
  • Debate continues over how much x86 decoding and legacy constraints hurt perf‑per‑watt versus ecosystem control and integration (e.g., Apple’s advantages).

Broader Tech Analogies & Article Reception

  • Comparisons are drawn to NeXT (technically influential but commercially weak) and IBM’s misplays (BIOS assumptions, MCA), as examples of strong players misreading markets.
  • Some feel the article ends abruptly around 2013 without covering the crucial downturn years, leaving the central “how Intel lost” question underexplored.